Control strategy for multipass robotic grinding

This paper describes a control strategy for removing unwanted material in multiple passes using a robotic disk-grinding system. The system, which is interlinked with a PC, consists of an industrial robot equipped with a pneumatic grinder, a six-degree-of-freedom force sensor, and two noncontact inductive proximity sensors. During grinding, a generalized predictive controller adapts its own decision following an optimal locus and controls the process in real time, achieving maximum metal removal rate under certain process and robot system constraints. The normal grinding force and robot feed speed are used as control parameters. The experimental results demonstrate the successful implementation of this grinding strategy and confirm that larger volumes of material can be optimally ground in repetitive several passes. Finally, a good surface profile with specified finish can also be achieved using a surface flattening algorithm.